VB33R

VB33R
Straus,
Straus, M.
M. A.
A. (2005).
(2005). Women's
Women's violence
violence toward
toward men
men isis aa serious
serious social
social problem.
problem. In
In D.R.
D R . Loseke,
Loseke, R.
R. J.
J. Gelles
Gelles &
& M.
M. M.
M
Cavanaugh
Cavanaugh (Eds.),
(Eds.),Current
Current controversies
controversies on
on family
famlly violence,
violence, 2nd Edltlon (2nd
2nd Edition (2nd Edition
Edition ed.,
ed., pp.
pp. 55-77).
55-77). Newbury
Newbury Park:
Park:
Sage
Sage Publications.
Publications.

Women's Violence
Toward Men is a
Serious Social Problem
Murray A. Straus

T h e first purpose of this chapter is to review research showing that

w o m e n initiate a n d carry o u t physical assaults o n their partners as
often as d o men. A second purpose is to show that, despite the m u c h

AUTHOR'S NOTE: This chapter is a revision and updating of a paper

presented at the 1989 meeting of the American Society of Criminology it is a
pleasure to acknowledge the comments and criticism of the members of the
1989-90 Family Research Laboratory Seminar, and also Angela Browne,
Glenda Kaufman Kantor, Coramae Mann, Daniel Saunders, Kirk R. Williams,
and Kersti A. Yllo. However, this does not imply their agreement with this
chapter. Part of the data are from the National Family Violence Resurvey,
funded by National Institute of Mental Health grant R01MH40027
(Richard J. Gelles and Murray A. Straus, co-investigators) by a grant for
"Family Violence Research Training" from the National Institute of Mental
Health (grant T32 MH15161).
 56 DEFINITION AND MEASUREMCNl CONTROVERSIES: WOMEN'SVIOLCNCE

lower probability of physical injury resulting from attacks by women,

women produce a substantial percentage of all injuries and fatalitiesfrom
partner violence.
"Minor" assaults perpetrated by women are also a major problem,
even when they do not result in injury, because they put women in
danger of much more severe retaliation by men. They also help per-
petuate the implicit cultural norms that make the marriage license a
hitting license (Straus & Hotaling, 1980).It will be argued that in order
to end "wife beating," it is essential for women also to end what many
regard as a "harmless" pattern of slapping, kicking, or throwing sorne-
thing at a male partner who persists in some outrageous behavior and
"won't listen to reason."
The chapter focuses on physical assaults, even though they arc not
necessarily the most damaging type of abuse. One can hurt a partner
deeply-even drive them to suicide-without ever lifting a finger.
Verbal aggression may be even more damaging than physical attacks
(Vising, Straus, Gelies, & Harrop, 1991). This chapter focuses exclu-
sively on physical assaults because, with rare exception, the contro-
versy has been focused on this type of violence by women. Detailed
methodological and sociology of science analyses of the controversy
can be found in Felson (2002) and Straus (1999).

f* DEFLNING AND MEASURING ASSAULT

The National Crime Panel Report defined nssnult as "an unlawful phys-
ical attack by one person upon another" (US. Department of Justice,
1976).It is important to note that neither this definition, nor the defini-
tion used for reporting assaults to the FBI (Federal Bureau of Investi-
gation, 1995), requires injury or bodily contact. Nevertheless, injury will
be considered in this chapter for two reasons. First, the presence of
injury makes a difference in what the police, prosecutors, and juries do.
Second, numerous studies show that a substantial proportion of serious
injuries and homicides of partners are perpctrated by women.

+*: GENDER DIFFERENCES

IN PARTNER ASSACLT AND HOMICIDE

Violence by women against male partners has been a difficult and

controvers~alissue caused by differences in research methodologies
 Women's Violencc Toward Men 57

and in moral agendas (Straus, 1999). One of the major discrepancies in

research is between what can be called "family conflict" studies, such
as the National Family Violence Surveys, and "crime studies," such as
the National Crime Victimization Survey Family conflict studies ask
respondents about problems and conflicts in their family, while crime
studies focus on examining police reports or asking respondents if they
have been victims of crime.
Witlxout exception, family conflict studies find approximately
equal rates of assaults by women and men (Archer,2000; Fiebert, 1997).
In comparison, and also without exception, crime studies find much
higher rates of assaults by men. Family conflict and crime studies also
yield extremely different answers to questions about the overall preva-
lence of assaults on partners: Crime studies find a fraction of the rates
found by family conflict studies. Both the low overall rate of assault
and the high percentage of assaults by men found in crime studies
probably occur because crime studies deal with only the small part of
all domestic assaults that study respondents experience as a "crime."
Assaults perceived as crimes rather than as "family fights" occur rela-
tively rarely and involve perpetration by men much more often than by
women (Straus, 1999).

Family Conflict Studies

Nntiorznl Fnniily Vivleizce Siiriieys. These studies have obtained data from
nationally representative samples of 2,113 married and cohabiting
couples in 1975 and 6,002 couples in 1985. In both surveys, the rate of
female-to-male assault was slightly higher than the rate of male-to-
female assault (Straus & Gelles, 1986, 1990). Because the seeming
equality in assault rates may occur because of a tendency by men to
underreport their own assaults (Dutton, 1988; Stets & Straus, 1990),the
assault rates were recomputed for this chapter on the basis of infor-
mation provided by the 2,994 women in the 1985 National Family
Violeuce Survey The resulting overall rate for assaults by women was
121per 1,000 couples, as compared to 122 per 1,000 for assaults by men
ns reporfed by their female p ~ ~ t i ~ eThis
r s . difference is not great enough to
be statistically reliable.
Separate rates were also computed for minor and severe assaults.
The rate of minor assaults by women was 78 per 1,000 couples, com-
pared wit11 a rate for men of 72 per 1,000. The severe assault rate
was 46 per 1,000 couples for assaults by women and 50 per 1,000 for
58 DEFINITION AND MEASUKCMENT CONTKOVERSIES: WOMEN'S VIOLENCE

assaults by men. Neither difference is statistically significant. Since

these rates are based exclusively on information provided by women
respondents, the near-equality in assault rates cannot be attributed to a
gender bias in reporting.

Other Funlily Violmce Szirzvys. There have been more than 100 family
violence surveys, whicli have used a variety of measures and reported
similar results. This includes research by respected scholars such
as Scanzoni (1978) and O'Leary, Malone, and Tyree (1994); and large-
scale studies such as the Los Angeles Epidemiology Catchment Area
study (Sorenson & Telles, 1991), the National Survey of Households
and Families (Brush, 1990), the Dunedin, New Zealand, birth cohort
study (Moffitt, Caspi, Rutter, & Silva, 2001), and a statewide survey
conducted for the Kentucky Commission on Women.
The Kentucky study raises a troublesome question of scientific
ethics, because it is one of several in which the data on assaults by
women were intentionally suppressed. The existence of that data
became knorzm only because FIornung, McCullough, and Sugimoto
(1981) obtained the computer tape and found that, among the violent
couples, 38 percent werr attacks by women on men who, as reported
by the women themselves, had not attacked them. More often, the
strategy to maintain the myth that partner assault is exclusively a male
crime has been to omit questions that ask about violence by women, as
for example in the Canadian National Survey of Violence against
Women.

Samples of "Bnttered Wmieiz." Studies of residents in shelters for

battered women are sometimes cited to show that it is only male part-
ners who are violent. However, these studies display the pattern of
deception and cover-up noted in the previous paragraph. They rarely
obtain or report information on assaults by women; and when they do,
they ask only about women's use of violence in self-defense.One of the
few exceptions is Walker (19841, who found that 1 out of 4 women in
battering relationsliips responded affirmatively when asked if she had
"used physical force to get something you wanted" (p. 174).Giles-Sims
(1983) also found that in the year prior to coming to a shelter, 50 percent
of the women reported assaulting their partner, and in the six months
after leaving the shelter, 41.7percent reported an assaiilt against a part-
ner. Giles-Sims's case study data suggest that is not likely these assaults
were in self-defense.
 Worncn's Violencc Toward Men 5Y

Dating Co~iples.Sugarman and Hotaling (1989) summarized the results

of 21 studies of violence in dating relationships. They found an average
assault rate of 329 per 1,000 for men and 393 per 1,000 for women; that
is, a higher proportion of females than males self-reported perpetrating
an assault on a dating partner. Other studies (Pirog-Good & Stets, 1989;
Stets & Straus, 1990) further confirm the equal or higher rate of assault
by women in dating relationships. The most extensive of these is the
International Dating Violence Study Preliminary results based on
research on more than 8,000 couples at 33 universities in 16 countries
show that the pattern of equal or higher rates of violence by women is
a worldwide phenomenon (Straus & Members of the International
Dating Violence Research Consortium, 2003).

Crime Studies
Niztionnl Crime Victimizntion S~irvey.Conducted for the Department of
Justice by the Bureau of the Census, the National Crime Victimization
Survey (NCVS) is an annual study of approximately 60,000 house-
holds. In comparison to family violence surveys, the NCVS finds a very
low prevalence rate of assault: fewer than 10 per 1,000 couples. The
NCVS rate for assaults by female partners was 11 per 1,000, and for
male partners 77 per 1,000. Thus, according to the NCVS, the rate of
domestic assaults by men is seven times greater than the rate of assault
by female partners.
The extremely low rate of assaults by both men and women found by
the NCVS may occur because the NCVS is presented to respondents as a
study of criiiie. The problem is that it takes relatively rare circumstances,
such as an injury or an attack by a former partner, to perceive an attack as
a "crime" (Langan & h e s , 1986).This is probably why the NCVS pro-
duces such totally implausible statistics such as a 75 percent injury rate
(compared with an injury rate of less than 3 percent in the family violence
surveys), and more assaults by fonner partners than by current partners.

Police Calls. Data on calls to the police about domestic assaults are
biased in ways that are similar to the bias of the National Crime Suuvey
Like the NCVS, at least 93 percent of the cases are missed (Kaufman
Kantor & Straus, 1990),probably because there was no injury or fear of
serious injury great enough to warrant calling the police. Since the
cases for which police are called tend to involve injury, or chronic
severe assault, and because that tends to be a male pattern, assaults by
60 DEFINITION AND MEASUREMENT CON~IROVERSIF5:WOMEN'S\IIOLENCE

women rarely arc recorded in police records. Another reason assaults

by women are rare in police statistics is that many men are reluctant to
involve the police (Fdson, 2002) and admit that they cannot "handle
their wife." These artifacts produce a rate of assaults by men illat is
hugely greater than the rate of assault by women.

Nnlional Violence Against Wonzeiz Survey (NVAW). Sponsored by the

National Institute of Justice and the Centers for Disease Control, the
NV.4W surveyed 8,000 women and 8,000 men representing 16,000
households (Tjaden & Thoennes, 2000). The initially released results
reported that men physically assaulted their female partners at three
times the rate at which women engaged in such behavior. This was
interpreted as evidence showing that domestic violence is a male
crime. There were, however, several problems with this widely dis-
seminated conclusion. First, although the rate of perpetration by men
was t h e e times greater, an unbiased interpretation would have also
noted that women committed a third of domestic assaults-one-third
of offenders cannot be ignored. Second, buried in publications released
a year later was a table giving the past-year prevalence rates, as con-
trasted with the lifctilne prevalence rates released earlier. Past-year
prevalence rates are the most usual way of reporting crime statistics,
and they are considered to be more accurate because they do not
depend on recall of events long past. When past-year prevalence rates
are used, women committed 39 percent of the partner assaults. Third,
the NVAW survey was presented to respondents as a study of crime
and personal safety, and therefore respondents were implicitly encour-
aged to restrict their reports to "real crimes," ilius excluding most
instances of assault by a parbier, and especially "harmless" assaults
by women. Thus, a study that, in my opinion, was carried out to refute
the idea of gender symmetry in partner vioience instead gave strong
support to the conclusion that women physically attack partners at
about the same rate as do men.

I'nl-iizer Homicide Rates. Homicides are widely believed to be the most

colnpletely recorded crime and therefore to be relatively free from the
reporting biases just described. Homicide rates published by t l ~ eFBI
show that only 14 percent of homicide offenders are women (Federal
Bureau of Investigation, 1988). However, the percentage of women
offenders varies tremendously according to the relationship between
offender and victim.
 Women's Violcncc Toward Mcn 61

Female-perpetrated homicides of strnngers occur at a rate that is less

than a twentieth the male rate. The female share goes up somewhat for
murders of ncqziniiztnnces.As for murders of fniizily n~einbe~s,
women commit
them at a rate that is almost half the rate of men in the period 1976-79 and
more than a third of the male rate during the period 1980-84. However,
"family" indudes all r e l a h ~ swhereas
, the main focus of this chapter is
couples. There are two gender-specific estimates of the rates for partner
homicides (Browne 8r Williams, 1989; Straus, 1986). These two studies
found that women murder male parhlers at rates that are 56 percent and
62 percent as great as the rate of partner homicides by men. This is far
from equality, but it indicates that, in partner relationsliips,even when the
assaults are so extreme as to result in death, a substantial proportion are
committed by women, whereas as noted previously, for murders of
strangers, the female rate is only a twentieth of the male rate.

'3 SIHOULD INJURY BE PART OF

THE DEFINITION OF PARTNER VIOLENCE?

As pointed out elsewhere (Straus, 1980), female assault rates based on

the Conflict Tactics Scales (CTS) used in the Family Violence Surveys
can be misleading if the study does not also examine the purpose of the
violence and the injuries resulting from assaults. The 1985 Natioi~al
Family Violence Survey included questions on who initiated violence
and cpstions on injuries. The revised CTS (Straus, Hainby, Boney-
McCoy, & Sugarman, 1996) includes supplemental questions on injury.

Iiljury-Adjust Rntcs. Stets and Straus (1990) and Brush (1990) provide
data that can be used to adjust the assault rates to take into account
whether or not the assault resulted in an injury. Stets and Straus found
a rate of 3 percent for injury-producing assaults by men and 0.4 percent
for injury-producing assaults by women. Somewhat lower injury rates
were found by Brush for another large national sample: 1.2percent for
injury-producing assaults by men and 0.2 percent for injury-producing
assaults by women. An "injury-adjusted" rate was computed using
the higher of the two injury estimates. The resulting rate of "injury-
producing assaults" by men is 3.7 per 1,000, and the rate of injury-
producing assaults by women is much lower: 0.6 per 1,000. Thus, the
injury-adjusted rate for assaults by men is six times greater than the
rate of domestic assaults by wonlei?.
 62 DEFINITION AND MEASUREMENT CONTROVERSIES: WOMEN'SVIOLCNCE

Although the injury-adjusted rates highlight the greater injury

inflicted by male offenders, there are several disadvantages to rates based
on injury (Straus, 1990,pp. 79-83). One of the disadvantages, for example,
is that the criterion of injury contradicts the domestic assault legislation
and new police policies, which are major achievements in the efforts
to end violence against women. These statutes and policies premise
restraining orders and encourage arrest on the basis of attacks. The
woman does not have to suffer an observable injury for action to be taken.
Another disadvantage of using injury as a criterion for domestic
assault is that injury-based rates omit the 97 percent of assaults by men
that do not result in injury but that are nonetheless a serious social
problem. Without an adjustment for injury, National Family Violence
Survey produces an estimate of more than 6 million women assaulted
by a male parhler each year, of which 1.8 million are "severe" assaults
(Straus & Gelles, 1990).If the injury-adjusted rate is used, the estimate
is reduced to 188,000 assaulted women per year. The figure of 1.8
million seriously assaulted women each year has been used in many
legislative hearings and countless feminist publications to indicate the
prcvaleiice of the problem. If that estimate was replaced by 188,000, it
would understate the extent of the problem and could handicap efforts
to educate the public and secure funding for shelters and other ser-
vices. Fortunately, that is not necessary. Both estimates can be used,
since each highlights a different aspect of the problem.

'f SELF-DEFENSE AND ASSAULTS BY WOMEN

For many years T explained the high rate of attacks on partners by

female partners as largely a response to or a defense against assault by
their partner. However, new evidence raises questions about that
interpretation.

Homicide
For lethal assaults by women, some studies suggest that a
substantial proportion are self-defense, retaliation, or acts of despera-
tion following years of brutal victimization (Browne, 1987; Browne &
Williams, 1989;Jurik & Gregware, 1989).However, Jurik and Gregware's
(1989) investigation of 24 cases in which women killed male partners
found that the victim initiated use of physical force in 40 percent of the
 Women's Violcnce Toward Mcn 63

cases, and that only 21 percent were in response to "prior abuse" or

"threat of abuse/death." They also found that 60 percent of the women
had a previous criminal record. Likewise, Mam's (1990) study of the
circumstances surrounding partner homicides by women shows that
many women who murder their partners are impulsive, violent, and
have criminal records.

National Family Violence Survey

Female-Only Violence. Of the 495 couples in the 1985 National Family
Violence Survey for whom one or more assaults were reported by a
woman respondent, the man was the only violent partner in 25.9 percent
of the cases; the female partner was the only one to be violent in 25.5
percent of the cases; and both were violent in 48.6 percent of the cases.
Thus, a minimum estimate of violence by women that is not self-
defense because she is the only one to have used violence in the past
12 months is 25 percent. Brush (1990) reports similar results for the
couples in the National Survey of Families and Households and the
National Comorbidity Study.
Perhaps the real gender difference occurs in assaults that are
severe enough to carry a high risk of causing an injury, such as punch-
ing, kicking, and attacks with weapons. This hypothesis was investi-
gated using the 211 women who reported one or more instances of a
"severe" assault. The resulting percentages were similar: Both used
violence in 35.2 percent, male only in 35.2 percent, and female only in
29.6 percent.
Regardless of whether the analysis is based on all assaults or is
focused on dangerous assa~dts,about as many women as men attacked
a partner who had no1 hit them during the one-year referent period.
This is inconsistent with the "self-defense" explanation for the high
rate of domestic assault by women. However, it is possible that, among
the couples where both assaulted, ail the women were acting iii self-
defense. Even if that unlikely assumption were correct, it would still
remain that 25-30 percent of violent relationships are violent solely
because of attacks by the female partner

Ii~iiinfioizof Altnclcs. The 1985 National Family Violence Survey asked

respondents, "Let's talk about the last time you and your partner got into
a physical fight a n d . . . (the most severe act previously mentioned) . . .
happened. In that particular instance, who started the physical conilict,
61 DEFINITION AND MCASUREMCNr CONTROVEIISIES: WOMEN'SVIOLENCE

you or your partner?" According to the 446 women involved in a violent

relationship, their partners struck the first blow in 42.3 percent of the
cases, they hit first in 53.1 percent of the cases, and they could not
remember or could not disentangle who hit first in the remaining
3.1 percent of the cases. Similar results were obtained by other studies
(Archer, 2000).

Is the High Rate of Assault

by Women Explainable as Self-Defense?
It is remarkable that when research does not preclude the possibil-
ity of women being the instigators of violence by omitting data on
female perpetrators, every study finds that women initiate violence
in a large proportion of cases. Let us assume that many of the assaults
initiated by women are in response to fear derived from a long prior
history of victimization. Even if that is the case, it is a response that
tends to elicit further assaults by male partners (Bowker, 1953; Feld &
Straus, 1989; Gelles & Straus, 1988, chap. 7; Straus, 1974) and therefore
helps to perpetuate or increase partner violence.

+:GENDEIZ
+ AND CHRONIClTY OF ASSAULT
Although the prevalence rate of assaults by women is about the same as
that for men, men may engage in more rcpcnted attacks. This hypothesis
was investigated by computing the mean number of assaults among
couples for which at least one assault was reported by a female res-
pondent. According to these 495 women, their partners averaged 7.2
assaults during the year, and they themselves averaged six assaults.
Although the frequency of assault by men is greater than the frequency
of assault by women, the difference is not large enough to be statistically
dependable. If the analysis is restricted to the 165 cases of severe assault,
the men averaged 6.1 and the women 4.3 assaults, whidi is a 42 percent
greater frequency of severe assault by men and is just short of being
statistically significant. If one disregards the tests of statistical signifi-
cance, these comparisons support the hypothesized greater chronicity
of violence by men. At the same time, the fact that the average number
of assaults by men is higher should not obscure the fact that the violent
ioomelz carried out an average of six minor and five severe assaults per
year, indicating a repetitive pattern by women as well as men.
 Women's Violence Toward hlcn 65

*+: COINTEXT, MEANING, AND MOTIVES

The symmetry between males and females in the number and severity
of assaults, important as it is, ignores the context, meaning, and con-
sequences of these assaults. Feminist scholars believe that there are
important differences between men and women in the motivation
for assaults on a partner. Howevel; less injury seems to be the only dif-
ference that has been well documented by empirical research. A few
studies suggest, but do not demonstrate, difierences in context, mean-
ing, or motives. For example, a meta-analysis of research on gender
differences in aggression by Eagly and Steffen (1986) found no overall
difference in aggression by men and women, but less aggression by
women if the act would produce harm to the target. From this, one can
infer that women are more reluctant to inflict injury. Greenblat (1983)
interpreted her data as showing that men typically hit or threaten to hit
in order to force some specific behavior on pain of injury, whereas
women typically slap a partner or pound on his chest as an expression
of outrage or in frustration from his having turned a deaf car toward
repeated attempts to discuss some critical issue. Despite the surface dif-
ference, both are uses of physical violence for coerciou. One of the very
few empirical studies to investigate the motives for partner violence by
women found that the predominant explanation offered by the women
in the study was to coerce the partners into doing something (Fiebert &
Gonzalez, 1997).A careful review of the research by Felson (2002) led to
the conclusion tliat there was no clear evidence indicating differences in
the context, meaning, and motives for assaults by male and female part-
ners. Moreover, even if there were differences in context, meanings, and
motives, that mwuld not indicate the absence of assault by women. Nor
would it refute the hypothesis that assaults by women help legitimize
male violence. Only empirical research can resolve tliat issue.

+*: FEMALE OFFENDERS CAUSE

SUBSTANTIAL INJUIW AND DEATH

Jt is important to realize that, although the rate of injury inflicted by

women is lower, it is a large enough proportion of the injuries and
deaths to be a severe social and public health problem by itself. Studies
have found that 12-40 percent of injuries and homicides are inflicted
by women. The NVAW survey found that women's violence led to
66 DEl~lNlTIONAND MEASUREMENT CONTROVERSIES: WOMCN'S VIOLENCE

40 percent of all the past year's injuries, created 27 percent of the

injuries requiring medical attention, and accounted for 38 percent of
the victims who lost time from work and 31 percent of the victims who
feared bodily injury (Tjaden & Thoennes, 2000). Other research has
found that women's violence resulted in 12 percent of assault-related
injuries requiring medical attention (Stets & Straus, 1990),50 percent of
injuries needing medical attention among a sample of high school
students (Molidor & Tolman, 1998), 40 percent of injuries suffered by
college student dating partners (Makepeace, 1989), and a third of all
homicides of domestic partners (Remison, 2000).
The fact that men inflict a larger percentage of the severe injuries
and deaths does not diminish that the proportion perpetrated by
women is a serious health, crime, and family problem.

*: VIOLENCE BY WOMEN 1NCREASES

THE PROBABILITY OF VIOLENCE AGAINST WOMEN

There seems to be an implicit cultural norm permitting or encouraging

minor assaults by women in certain circumstances. Stark and McEvoy
(1970) found about equal support for a wife hitting a husband as for
a husband hitting a wife; Greenblat (1983) found that both men and
women are inore accepting of women hitting husbands than of husbands
hitting wives, and she suggests this is because fernale aggressors are
far less likely to do physical harm. These norms tolerating low-level
violence by women are transmitted and learned in many ways. For
example, even casual observation of the mass media suggests that just
about every day, there are scenes depicting a man who makes an insult-
ing or outrageous statement and an indignant woman who responds
by "slapping the cad." This presents an implicit model of assault as a
morally correct behavior to millions of women.
Although the previous section of this chapter demonstrated that
women are responsible for an important proportion of serious injuries
and deaths of partners, 1 assume that most of the assaults by women
fall into the "slap the cad" genre and are not intended to, and only
rarely cause, physical injury The danger to women is shown by stud-
ies finding that minor violence by women increases the probability of
severe assaults by men (Bowker, 1983; Feld & Straus, 1989; Gelles &
Straus, 1988, pp. 146-156). Sometimes this is immediate and severe
retaliation. But regardless of whether that occurs, a more indirect and
 Women'iViolenceToward Men 67

probably more important effect may occur because such morally correct
slapping acts out and reinforces the traditional tolerance of assault in
marriage. The moral justification of assault implicit when a woman
slaps or throws something at a partner for something outrageous
reinforces the moral justification for slapping her when she is doing
something outrageous, being obstinate, nasty, or "not listening to rea-
son" as he sees it. To the extent that this is correct, one of the many
steps needed for primary prevention of assaults on women is for
women to forsake even "harmless" physical attacks on male partners
and children. Women must insist on nonviolence by their sisters, just as
they rightfully insist on it for men.
It is painful to recognize the high rate of domestic assaults by
women. Moreover, the statistics are likely to be used by misogynists
and apologists for male violence. My view of recognizing violence by
women is parallel to Hart's (1986, p. 10) view on the importance of
recognizing battering within lesbian relationships. It is painful, but
to do otherwise obstructs a potentially important means of reducing
assaults by men-raising the consciousness of women about the
i~nplicitnorms that are reinforced by a ritualized slap for outrageous
behavior on the part of their partners.
It follo~vsfrom this discussion that efforts to prevent assaults by men
must also include attention to assaults by women. Although this may
seem like "victim blaming," there is an important difference: Recognizing
that violence by women is one of the many causes of violence against
women does not justify violence by men. It is the responsibility of men
as well as women to refrain from physical attacks (including retaliation),
at home as elsewhere, no matter what the provocation.

+:+
GENDER DIFFERENCES
IN TRENDS IN PARTNER VIOLENCE

The acceptability of hitting a partner and the actual rate of partner

violence in the United States has been decreasing in the past 25 years.
This decrease has been primarily in violence by male partners (Straus,
1995; Straus & Gelles, 1986; Straus & Kaufman Kantor, 1997).Yet despite
the decrease, partner violence by both men and women remains the
most frequent form of interpersonal violence in American society
The fact that violence and approval of violence by male partners
has decreased, whereas violence and approval of violence by female
 68 DFi:iNITiON AND AIEASUREMENT CONTROVERSICS: WOMEN'S VIOLENCE

partners has not, may reflect the fact that almost all programs to end
partner violence were created by and continue to be a major effort of
the women's movement. Comequently, they are based on the assumption
that partner violence is perpetrated almost exclusively by men. The
voluminous research summarized in this chapter shows that this
assumption is false. Most partner violence is mutual. Therefore, as
indicated previously, rather than ignoring assaults by female partners,
primary prevention of violence against women requires strong efforts
to end assaults tn) women. However, the needed change must be made
with extreme care. First, it must be done in ways that simultaneously
refute the idea that violence by women justifies or excuses violence
by their partners. Second, although women may assault partners at
approximately the same rate as men, assaults by meu usually inflict
greater physical, financial, and emotional injury This means that male
violence against women is typically the more serious crime. Tlius,
major focus on violence by women does not necessarily mean equal
iocus. Finally, in many societies women lack full economic, social,
political, and human rights. In such cultural contexts, equality for
women needs to be given priority as an even move fundanlental aspect
of primary prevention. Otherwise, focusing on partner violence by
women call further exacerbate the oppression of women.

*> CONCLUSIONS

Ending assaults by women needs to be added to efforts to prevent

assaults oiz women for a number of reasons. Perl~apsthe most funda-
mental reason is the intrinsic moral wrong of assaulting a partner. A
second reason is the fact that women inflict a third of theinjuries and
deaths from partner violence. Third, women who hit their partners
"model" violence for children, and this is associated with an increase
in psychological problems of children. The harm to children from
assaults by women is at least as strong as from assaults by men
(Iiolden, Geffner, Pr Jouriles, 1998; Jaffe, Wolfe, & Wilson, 1990; Straus,
1991). Fourth is the danger of escalation when women engage in
"harmless" minor violence. Feld and Straus (1989) found that if the
female partner also engaged in an assault, it increased the probability
that assaults will persist or escalate in severity over the one-year period
of their siudy; whereas if only one partner engaged in physical attacks,
the probability of cessation increased. Finally, when women assault
 Women's Violence Towaid Mcn 69

their partners, it validates the traditional cultural norms tolerating a

certak level of violence between parhlers and therefore helps perpetuate
a system in which they are the predominant victims.
It should be emphasized that the preventive effect of reducing
violenceby women, including "harmless" minor violence, has not been
demonstrated by the evidence in this chapter. It is a plausible inference
and a hypotheses for further research. However, it is important not to
wait for the results of such research before implementing steps to end
partner violence by women because, as pointed out, it would be equiv-
alent to ignoring the legal and moral wrong of such behavior, and
ignoring the physical and psychological injuries to their partners and
children. The steps can include posters and public service announce-
ments, police arrest policies, treatment programs for female offenders,
and school-based prevention programs addressed to girls as well as to
boys (Foshee, 2004). These steps must be made with extreme care for
a number of reasons, not the least of which is to avoid implying that
violence by women justifies or excuses violence by their partners.
Moreover, although women may assault their partners at approxi-
mately the same rate as men, the first priority in services for victims
and in prevention and control must continue to be directed toward
assaults by men because these tend to result in greater physical, fina11-
cial, and emotional injury.

*: RESPONSE TO I.OSEKE AND KURZ

The objections that Loseke and Kurz (this volume) raise to my chapter
reflect three major differences between us: theoretical differences,
methodological differences, and differences in our moral agendas.

Theoretical Differences
The theoretical difference is epitomized in a single word in the
titles to our chapters. My chapter refers to violence by women as "a"
social problem, wheveas their chapter asserts that violence against
women is "the" social problem. I do not believe that either violence by
men or by women is "the" problem. Society faces multiple and inter-
related problems with violence, and the correction of one usually
depends on dealing with the configuration of problems in which it is
embedded. Thus, violence against women is 11 serious social problem,
 70 DEFINITION AND MEASUREMENT CONTROVERSIES: WOMEN? VIOLENCE

but it is also only one aspect of tlie problem of violence in American,

and many other, societies. From the theoretical perspectives that social
problems are interrelated, and that violence is a multiply determined
interactive event, an adequate solution to the problem of violence against
women requires addressing the behavior of 'both participants in that
interactive sequence, as well aa addressing many other phenomena
that increase the risk of violence.
The sii~gle-piobleni focus epitomized in Loseke and Kurz's title is
part of a larger tlieoretical difference: a single-cause theoretical approach.
A single-cause approach has long been rejected by social scientists.
One exception, however, is the subgroup of feminist social scientists
who assume that a patriarchal social system and male dominance and
privilege explains almost all cases of violence against women.
It is appropriate and necessary for feminist scholars to focus their
research on this one of tlie many causes of violence against women.
Society is indebted to the feminists for bringing genderhased oppression
and violei~ceto the fore. At the same time, denying the importance of other
causes of violence against women, such as stress, alcoholisn~,violent
socialization, criminal propensities, and violence by womm, is something
that would be ridiculed and rejected if it came from social scientists of any
other tl~eoreticalpersuasion. The following section suggests why we
accept this scientific error from feminists but not from others.

The Moral Agenda

One of the reasons social scientists and the public at large are
willing to accept a single-cause approach adsocated by feminists is tlie
recognition of and indignation by most social scientists over past and
continuing oppression and discrimination against women. As a result,
there is a tendency to accept almost anything that will change this
aspect of society Liberal social scientists tend deliberately to close their
eyes to excesses and incorrect statements by feminists because they
do not want to undermine feminist efforts to bring about a more equi-
table society. Thus, avowedly feminist scholars have suppressed data
on violence by women. Social scientists I know who do not claim to be
feminists have also published only the part of their data that shows
violence by men.
History is full of atrocities carried out in the service of a moral
agenda. These make suppression and denial of evidence on feinale aio-
lence trivial by comparison. But to those like myself, tor whom ending
 Women's Vinlcnce Toward Men 71

1711violence, from spanking by parents to nations engaging in war, it is

as essential to confront violence by women against male partners as it
is to confront the high rate of violence by men.
Both Loseke and Kurz and 1 are against all violence, and both they
and 1 are against all forms of gender inequality The difference between
us is in priorities for researcli and action. I infer that they rank ending
oppression of women as number 1. Ending all violence is also very
important, but not number 1. On the other hand, I rank ending all
violence as number 3 and ending oppression of women as also very
important, but not number 1. They are willing to accept certain costs to
achieve equality for women, and I am willing to accept certain costs to
achieve a nonviolent society For example, although domestic violence
victims who need the services of a shelter are overwhelmingly women,
I am willing to accept the cost of radical male advocacy groups misus-
ing the results of my research to oppose shelters for domestic violence
victims that do not provide the same services for male victims. 1 am
willing to accept the rare instances in which they have been successful
as a bearable cost, because there is no way of avoiding it without sup-
pressing the evidence on female violence.
Violence by both men and wolnen against a partner are criminal
acts and moraily repulsive, except in the rare cases of self-defense. The
moral priorities of Loseke and Kurz represent a legitimate difference in
assessing the long-run costs and benefits for women of recognizi.ng that
women assault their parhlers at about the same rate as men. 1 think my
moral priorities promise a greater long-run benefit to women b-. ~iause,
for the reasons given in my chapter, ending violence b,y women will help
end violence n ~ n i l l s twomen. In addition to the other reasons in my
chapter explaining why ending violence by women will help end vio-
lence against women, I sl~ouldhave pointed out that it will end the
training in violence of the next generation of both men and women that
is provided when children grow up watching their mothers hit their
fathers. As my chapter shows, this occurs just as often as fathers hitting
mothers, and mothers are the first to hit as often as fathers.

Methodological Differences
One key methodological difference is that I believe that feminist
research, like all other research, cannot be limited to ill-depth qualita-
tive studies. Quaiitative studies are essential, but so are large-scale
surveys. Each has its own limitations, and each has the power to shed
 72 DEFINITION AND MEASUREMENT CONTROVERSIES: WOMEN'S \/!OI.CNCC

light 011 a differentaspect of violence between partners. Moreover, there

is also something not quite appropriate when Loseke and Kurz reject
the results of large-scale surveys showing symmetry in physical assault,
but accept the results of large-scale surveys showing a predominance
of male perpetrators in respect to sexual coercion.
As for the purported deficiencies in the CTS mentioned by Loseke
and Kurz, none is correct. The space I have been given for this rejoin-
der does not permit me to respoud to each of these purported defi-
ciencies, so I respond to just the first of them. Readers can find the
others on my Web site, http://pubpages.unh.edu/-1nas2,by clicking on
"Detailed Response to Loseke and Kurz."
Loseke and Kun. argue that "research based on representative
samples [using the CTS] will ~iildevesiiinatethe amount of extreme
violence experienced by women because severely abused women will
not participate in the survey" (italics in original). That is certainly true,
but it is even more likely that male victims of female violence avoid
participation in such surveys. Assaulting or being assaulted by a part-
ner is shameful. It took a niajor and still continuing effort by feminists
to get women to report such assaults to police. The same shame and
reluctance to participate in surveys occurs for male victims. Ilowever,
for men, there is the additional shame and reluctance stemmi~igfrom
the type of masculinity that expects a "real nian" to bc able to handle
such situations, and that lead police to scoff at or laugh a t inen who do
rile a complaint (Mills, 2003).

Other Inaccuracies in the Loseke-Kurz Article

There are a large number of othcr incorrect statements in the
Loseke-Kurz chapter. As in the case of the erroneous deficiencies of the
CTS,the space available to me permits including only the first two of
them. The others are on my Web site.
Loseke and Kurz claim that l "trivialize . . . the complex meaning
of violence and its impact on the lives of women." Their dernonstra-
tion of this, however, is a statement out of context, which reverses its
meaning. They say I characterize "women's t>yical violence as moti-
vated by their desire to 'slap the cad."' On the contrary, the "slap the
c a d phrase was not to show that this is typical. It was in a section of
my chapter designed to show that even such trivial violence increases
the risk of being attacked by a male partner. This is the opposite of
failing to recognize the impact of violence on the lives of women.
 \%men's Violence Toward Men 73

Moreover, the sentence preceding pointed out that "the previous

section. . . demonstrated that women are responsible for an important
proportion oC serious injuries and deaths of partners." This is not
trivial violence.
Loseke and Kurz say that I do "not incorporate gender at the level
of measurement." This is the opposite of what the record of my
research shows. For example, the Conflict Tactics Scales (CTS) was
designed to investigate gender differences in partner violence. In order
to show that there is more violence by male than female partners, the
CTS asks about assaults by both partners. My intention to show greater
male violence was thwarted by the results. Other examples abound.
I carried out the first empirical study of partner violence to ineasuip (not
just talk about) feminist concepts such as inequality in power and
resources and social norms tolerating partner violence (for example,
Straus, 1976).
In co~iclusion,1 have ahvays believed and acted on the belief
that a feminist approach is both valid and necessary. By a "feminist
approach," I mean taking into account phenomena that represent
gendered inequality and oppression. That is why 1 have taken that
approach in much of the research just cited. Feminist advocacy is
needed and is critical to free society of its sexist structure. But it has
gone beyond stimulating and motivating research to self-censorship
and attempts to suppress the results of other researchers whose find-
ings d o not co~iformto the feminist assumption that only men assault
partners. It undermined feminist credibility not just among researchers,
but also among the general public. That is tragic.
It is necessary to recognize without delay and to alert women to
the fact that violence against partners by is prevalent and is one
of the many causes of violence against women, just as violence by men
is prevalent and is one of the many causes of violence by women. There
is a difference between explanation and blame. The fact that violence
by women is part of the interactive sequence of events that constitutes
most partner violence does not excuse men any more than it excuses
women. It important to recognize this fact, primarily for the protection
of women but also to protect the reputation of feminist scholarship. I t
is important for the protection of women because each cause that is
identified provides an opportunity to develop programs to eliminate
or reduce that cause, and therefore to reduce partner violence. Each
cause that is identified and acted on adds to the effectiveness of the
effort to prevent violence against women.
74 DEFINITION AND MEASUIZEMENT CONTROVERSIES: WOMEN'SVIOLENCE

+*: REFERENCES

Archer, J. (2000). Sex differences in aggression between heterosexual partners:

A meta-analytic review. Psycliologicol Bulletin, 126(5),651-680.
Bowkcr, L. H. (1983). Beating uvjle bentiilg. kxington, MA: Lexington Books.
Browne, A. (1987). W!wn battered uwonierz kill. New York: Free Press.
Browne, A,,& Williams, K. R. (1989). Exploring the effect of resource avail-
ability and the likelihood of female-pelpetrated homicides. Law m d Society
Reoiczo, 23(1), 75-94.
Brush, L. D. (1990). Violent acts and injurious outcomes in married couples:
Methodological issues in the National Suuvey of Families and Households.
Gender. and Society, 4(1), 56-67.
Dutton, D. G. (1988). Profiling of wife assaulters: Preliminary evidence for a
trimodal analysis. Violnicc and Victims, 3(1), 5-29.
Eagly, A. H., & Steffen, V. J. (1986). Gender and aggressive behavior: A meta-
analytic review of the social psychological literature. Ps!jchological B~illetin,
100(3), 309-330.
Federal Bureau of Investigation. (1988). Criiiie in the Unlteti Stoies: Uit@~r!r!crinie
reporis. Washington, DC: U.S. Government Printing Office.
Federal Bureau of investigation. (1995). Crime in the United Stntcs, 7993: Uriiforni
ci-ime rcpoi.is: Washington, DC: U.S. Deparlment of Justice.
Fcld, S. L., & Straus, M. A. (1989). Escalation and dr?sistancc of wife assault in
mai-riage. Crlriiiiinlrigy, 77(1), 141-161.
Felson, R. B. (2002). Vloleizce mid gcildn- reexniiiiiid. Washington, DC: American
Psycliological Press.
Fiebert, M. 5. (1997). Ailnotated bibliography: Refcrcnces examining assaults
by women on their spouses/partners. In B. M. Dank & 11. Refinette (Eds.),
Scxlinl ho~nssiiic.iztniid ser~inlcoiisei~t(Vol. 1, pp. 273-286). New Brunswick,
NJ: Transaction.
Fiebert, M. S., & Gonzalez, D. M. (1997). College women who initiate assaults
on their niale partners and the reasons offered for such behavior.
Psycho!ogic[il liepork, 80, 583-590.
Fosher?,\<A. (2004). Siife dntiis: Aii oiio!esci.nt dntii~gnbiise preoentioil c~ir.r.icul~i~~z.
Centcr City, MN. Hazeiden Publishing and Educational Services.
Gelles, R. J., & Straus, M. A. (1988). Intiiiinte violence. New York: Simon &
Schuster.
Giles-Sims, J. (1983). Wife Imttei-ing: A systems t11eor.y npproncli. New York: Guilford.
Grcenblat, C. S. (1983) A hit is a hit is a h i t . . . o r is it? Approval and tolerance
of the use of physical force by spouses. In D. Finkellior, R. J. Gelles,
G. T. Hotaling, iYr M. A. Straus (Eds.), The dark side ojjii~niiies(pp. 235-260).
Beverly Hills, CA: Sage.
Hart, B. (1986). Preface. In K. Lobe1 (Ed.), Naming the violence: Speaking out about
lesbion battering (pp. 9-16). Seattle, MIA: Seal.
 Women's Violence Towaid Men 75

Holden, G. W , Geffner, R., & Jouriles, E. N. (Eds.). (1998). Chiidren exposed

to maritni ziiolence: Tlzeory, reseorcli, and upplied iss~ies.Washington, DC.
American Psychological Association.
Horn~mg,C. A,, McCullough, B. C., & Sugimoto, T. (1981). Status I-elationships
in marriage: Risk factors in spouse abuse. lo~rrrzdo,fMni-rigenrzd theFniiiily,
43,675-692.
laffe, P. G., Wolfe, D. A., & Wilson, S. K. (1990). Chiidwn of bnftered monien.
Newbnry Park, CA: Sage.
Jurik, N. C., & Gregware, P. (1989). A n~diiodfor mirds.: An inti'rnctioi?istonaly-
sis ofliomicides by wo~iieii.'Tempe:Arizona State University, School of Justice
Studies.
K a u h a n Kantor, G., & Straus, M. A. (1990). Response of victims and the police
to assaults on wives. In M. A. Strans & R. J. Gelles (Eds.), Pliysiurl violei?ce
in Aiiiri-icanfamilies: Risk factors cind ndnptations to oiolence in 8,145 fnmilies
(pp. 473387). New Brunswick, NJ: Transaction.
l.angan, P. A,, & Irmes, C. A. (1986). Preventing doiiirstic violence flgainst woirieir.
Washington, DC: U.S. Department of Justice, Bureau of Justice Statistics,
Special Report NCJ 102037.
Makepeace, J. M. (1989). Dating, living together, and courtship violence. 111
M. A. Pirog-Good & J. E. Stets (Eds.), Vioieiice in dntiny ipintionsliips
(pp. 94-107). New York: Praegei
Mann, C. R. (1990). Black female homicide in tlhe United States. Journal of
irite~l~crmo""iioiencc,5, 176-201
Mills, L. (2003). Ii~stilt to injur!~: Rcthinkiirg uiii. r e s p o ~ ~ sto~ sii~tinintenbuse.
Princeton, NJ: Princeton University Press.
Molfitt, T. E., Caspi, A., Ilutter, M., & Siiva, P. A. (2001). Sex d@i~i?cesin airtiso-
cini bcliniior~Cambridge: Cambridge University Press.
Molidor, C., & Tolman, R. M. (1998). Gender and contextual factors in adoles-
cent dating vioience. Vioioicp npitist Women, 4(2), 180-194.
O'Leary, I<. D., Malone, J., & Tpree, A. (1991). Physical aggression in early
marriage: Prerelationship and relationship effects. lo~i;,iniof Co;~sciltinyand
Cliiricnl Psychology, 62(3),5 9 G 0 2 .
Pirog-Good, M. A., & Stets, J. E. (1989). Viol~~im iri dztiny rrlntioirsliips: Emerging
social issiies. New York: Praegcr.
Remison, C. M. (2000). Irztiri~ntepnrtner uioln7ce. Wiashington, DC: Bureau of
Justice Statistics, Special Report NCJ 178247.
Scanzoni, J. (1978). Sex roles, woiirrn's work, and ninritni coizjlict. Lexington, MA:
Lexington Books.
Sorenson, S. B., & Telles, C. A. (1991). Self-reports of spousal violence in
a Mexican-American and non-Hispanic Wliite population. Violei~cenliii
Viciiirrs, 6(1), 3-15.
Stark, R., & McEvop Ill, J. (1970). Middle-class violence. Psycirology To~iaiy.4,
52-65.
 Women's Violenceroward Men 77

Straus, M. A., & Members of thc international Dating Violence Research

Consortium. (2004). Prevalence of violence against dating partners by
male and female university students worldwide. Violence Againsf Vv'omei!,
10(7), 790-811.
Sugarmen, D. B., & Hotaling, G. T (1989).Dating violence: Prevalence, context,
and risk markers. ln A. A. Pirog-Good and J. E. Stets (Eds.), Violence in dat-
ing relationships: Emerging socinl issues (pp. 3-31). New York: Praeger.
Tjaden, P., & Thoennes, N. (2000). Full report o f f h e prevalence, iircidence, and co!!-
sequences of violence ngainst wornel!: Findingsfiom the Natioizd Violence agniiisf
Women Survey. Washington, DC: US. Department of Justice, Office of
Justice programs, NCJ 183781.
US. Department of Justice. (1976). Dicfioimry of criinii~nljustice data terniinology.
Washington, DC: National Criminal Justice Information Service.
Vissing, Y M., Straus, M. A,, Gelhs, R. I., & Harrop,!. W. (1?91). Verbal aggres-
sion by parents and psychosocial problems of children. Child Abuse arid
Neglect, 15, 223-238.
Walker, L. E. (1984). The buffered ruulnan syndrunie. New York: Springer.